[RSS 2025] Learning to Act Anywhere with Task-centric Latent Actions
Python · active 2025-05-09 → 2026-07-09 (UTC)
Activity over time
Daily event counts in the loaded window
Line chart, 427 days from 2025-05-09 to 2026-07-09. Pushes: 23 total, peak 3 in a day. Pull requests: 3 total, peak 2 in a day. Issues: 66 total, peak 4 in a day. Comments: 103 total, peak 6 in a day. Stars: 481 total, peak 28 in a day.
- Pushes
- Pull requests
- Issues
- Comments
- Stars
Stars, PRs, issues and forks are under-captured in the later part of this window. GH Archive progressively stopped capturing non-push events during 2026 — −95% or worse by the end of the window. Every series here except Pushes fades for that reason, so a decline above reflects the archive, not this repository. Pushes stay reliable throughout, so read them, and the contributor counts derived from them, as the real signal. Data health has the measurements.
Top contributors
Pushes, PRs, issues, reviews and comments — stars and forks excluded, so this is contribution rather than popularity
| Contributor | Contributions | Pushes | PRs | Comments |
|---|---|---|---|---|
| retsuh-bqw | 79 | 17 | 0 | 48 |
| JackHuang0701 | 15 | 0 | 0 | 5 |
| minghaoguo20 | 7 | 0 | 0 | 4 |
| Leo-Yuyang | 6 | 0 | 0 | 5 |
| firefly-insky | 5 | 0 | 0 | 2 |
| lucasjinreal | 4 | 0 | 0 | 2 |
| NNNatsuki | 4 | 0 | 0 | 1 |
| Luo-Zhongwei | 4 | 0 | 0 | 3 |
| DuckSQ-BIT | 4 | 0 | 0 | 2 |
| serene-sivy | 3 | 2 | 0 | 1 |
| swstbecrpgmail | 3 | 0 | 0 | 1 |
| dreamcubeblock | 3 | 0 | 0 | 2 |
| feiyu12138 | 3 | 0 | 0 | 1 |
| bailvwangzi | 3 | 0 | 0 | 2 |
| luzhoulz | 3 | 0 | 0 | 1 |
| dajiebi | 3 | 0 | 0 | 1 |
| BarneetWei | 2 | 0 | 0 | 1 |
| Nemo-1024 | 2 | 0 | 0 | 2 |
| Freq-q | 2 | 0 | 0 | 2 |
| ginwind | 2 | 0 | 0 | 1 |
Recent activity
Latest issues, pull requests and releases
- Issue#7525616812442026-01-09 02:52self.dino_encoder = torch.hub.load('facebookresearch/dinov2', 'dinov2_vitb14_reg')
- Issue comment#7421Broccoli2026-01-06 14:01StageB finetune
- Issue comment#56Hitjss2025-12-24 09:01Question about CALVIN finetuning
- Issue#72xuxiaoxxxx2025-12-08 06:52The confusion of stage1 in pretraining
- Issue comment#72retsuh-bqw2025-12-05 09:31The confusion of stage1 in pretraining
- Issue#72xuxiaoxxxx2025-11-21 11:48The confusion of stage1 in pretraining
- Pull request#71OptimistiCompound2025-11-21 06:49
- Pull request#71OptimistiCompound2025-11-21 06:49
- Issue comment#60retsuh-bqw2025-11-19 02:57Loss instability when resuming pretraining from converted HF weights
- Issue comment#60Nemo-10242025-11-18 05:32Loss instability when resuming pretraining from converted HF weights
- Issue comment#65retsuh-bqw2025-10-02 06:37About exploiting wrist view camera
- Issue#60minghaoguo202025-09-23 10:43Loss instability when resuming pretraining from converted HF weights
- Issue comment#59ginwind2025-09-11 08:49Question about the python version of evaluating LIBERO
- Issue comment#16Freq-q2025-09-11 08:27Error when calling get_latent_action function (During Libero evaluation)
- Issue comment#59Freq-q2025-09-10 06:36Question about the python version of evaluating LIBERO
- Issue#63retsuh-bqw2025-09-08 11:10Question: Why is T hard-coded as 2 in rearrange operation when num_frames can be different?
- Issue comment#63retsuh-bqw2025-09-05 08:43Question: Why is T hard-coded as 2 in rearrange operation when num_frames can be different?
- Issue#63sigongzi2025-09-05 08:35Question: Why is T hard-coded as 2 in rearrange operation when num_frames can be different?
- Issue comment#62retsuh-bqw2025-09-04 08:18train loss and latent_action_accuracy when deploy univla to real world
- Issue comment#62BarneetWei2025-09-04 07:42train loss and latent_action_accuracy when deploy univla to real world
- Issue#62BarneetWei2025-09-04 06:58train loss and latent_action_accuracy when deploy univla to real world
- Issue comment#61feiyu121382025-09-02 04:22Training task-centric codebook (LAM stage-2)
- Issue#61feiyu121382025-09-02 04:22Training task-centric codebook (LAM stage-2)
- Issue comment#58retsuh-bqw2025-09-02 04:13training metric train/code_usage_step
- Issue comment#61retsuh-bqw2025-09-02 03:39Training task-centric codebook (LAM stage-2)
Totals cover only the window loaded into ClickHouse and count events, not GitHub's lifetime totals — 481 stars here means stars gained during the window, not the repo's star count.